Timezone: »

DepthShrinker: A New Compression Paradigm Towards Boosting Real-Hardware Efficiency of Compact Neural Networks
Yonggan Fu · Haichuan Yang · Jiayi Yuan · Meng Li · Cheng Wan · Raghuraman Krishnamoorthi · Vikas Chandra · Yingyan Lin

Tue Jul 19 10:40 AM -- 10:45 AM (PDT) @ Ballroom 1 & 2

Efficient deep neural network (DNN) models equipped with compact operators (e.g., depthwise convolutions) have shown great potential in reducing DNNs' theoretical complexity (e.g., the total number of weights/operations) while maintaining a decent model accuracy. However, existing efficient DNNs are still limited in fulfilling their promise in boosting real-hardware efficiency, due to their commonly adopted compact operators' low hardware utilization. In this work, we open up a new compression paradigm for developing real-hardware efficient DNNs, leading to boosted hardware efficiency while maintaining model accuracy. Interestingly, we observe that while some DNN layers' activation functions help DNNs' training optimization and achievable accuracy, they can be properly removed after training without compromising the model accuracy. Inspired by this observation, we propose a framework dubbed DepthShrinker, which develops hardware-friendly compact networks via shrinking the basic building blocks of existing efficient DNNs that feature irregular computation patterns into dense ones with much improved hardware utilization and thus real-hardware efficiency. Excitingly, our DepthShrinker framework delivers hardware-friendly compact networks that outperform both state-of-the-art efficient DNNs and compression techniques. All codes and pretrained models will be released upon acceptance.

Author Information

Yonggan Fu (Rice University)
Haichuan Yang (Facebook)
Jiayi Yuan (Rice University)
Meng Li (Facebook Inc)
Cheng Wan (Rice University)
Raghuraman Krishnamoorthi (Facebook)
Vikas Chandra (Facebook)
Yingyan Lin (Rice University)

Related Events (a corresponding poster, oral, or spotlight)

More from the Same Authors